Impact of SSH-tunnels on HTTP-traffic

For a project at work, I did some benchmark tests of the impact of SSH-tunnels on HTTP-traffic. And as we should suspect is true, on this test the traffic slows down with a factor 6. I did this, because some off-shore company out of India came up with some bad results from a benchmark, but they used SSH-tunnels.

So time for some testing... I did 6 types of tests using ApacheBench:

1. 10.000 requests with 200 concurrent connections without HTTP-keepalive via direct request to
   port 80 on the server
2. 10.000 requests with 200 concurrent connections with HTTP-keepalive via direct request to
   port 80 on the server
3. 10.000 requests with 200 concurrent connections without HTTP-keepalive via SSH-tunnel to
   port 80 on the server
4. 10.000 requests with 200 concurrent connections with HTTP-keepalive via SSH-tunnel to port
   80 on the server
5. 10.000 requests with 200 concurrent connections without HTTP-keepalive via SSH-tunnel with
   compression to port 80 on the server
6. 10.000 requests with 200 concurrent connections with HTTP-keepalive via SSH-tunnel with
   compression to port 80 on the server



For details check the report.




Phase Short Descr                         # Request/sec  Avg tim per request 
===== =================================== ============== ===================

   01 Direct                                     6323.94            0.158 ms
   02 Direct + Keepalive                         6319.27            0.158 ms
   03 SSH                                        1047.83            0.954 ms
   04 SSH + Keepalive                            1138.74            0.878 ms
   05 SSH + Compression                           824.09            1.213 ms
   06 SSH + Compression + Keepalive              1012.70            0.987 ms
===== =================================== ============== ===================